Anthropic's Claude can now orchestrate up to 1,000 AI agents in parallel through dynamic workflows
decoder ·
Anthropic has introduced dynamic workflows to its Claude Managed Agents, enabling the orchestration of up to 1,000 AI agents in parallel. This new infrastructure allows a lead agent to create plans, distribute tasks to sub-agents, and merge their results upon completion. While the cost-effectiveness of such large-scale agent swarms is debated, Anthropic cites internal testing where this method consistently identified 66 bugs in a large codebase compared to fewer findings by single agents. Despite these efficiency gains, the company warns that these workflows consume significant token resources. Consequently, users are advised to start with small-scale implementations to manage costs effectively. Activation requires selecting the specific "multiagent_20261001" agent type via documentation or the Claude Code interface.
decoder